Based on FY-3C VIRR LST and FY-3D/MWRI L1B brightness temperature data of February 1, 2020 and taking the area with geographical coordinates of 18°~54°N, 73°~135°E as an example, the LST reversion and downscaling based on the FY-3D/MWRI L1B data were studied using a statistical regression model and a hierarchical Bayesian fusion model. As a result, two models were constructed, namely a LST binary linear regression inversion model based on FY-3D single-channel horizontal and vertical polarization-corrected brightness temperature data and a hierarchical Bayesian fusion downscaling model based on FY-3D retrieved LST and FY-3C VIRR LST. They were verified with the LST on the day of MYD11A1 as reference data, obtaining the following results. As for the reversion statistical model, the mean bias, error standard deviation, and root mean square error were -1.28 K, 8.85 K, and 8.85 K, respectively for the FY-3D descending data and were -0.81 K, 6.74 K, and 6.78 K, respectively for the FY-3D ascending data. As for the hierarchical Bayesian fusion downscaling model, the mean bias, error standard deviation, and root mean square error were 0.50 K, 5.45 K, and 5.41 K, respectively for the FY-3D descending data and were 0.25 K, 5.54 K, and 5.54 K, respectively for the FY-3D ascending data. This study will provide a novel idea for the LST inversion and downscaling of passive microwaves.